Academic Document Q&A Agent

An agentic RAG (Retrieval-Augmented Generation) system that answers questions over academic documents using intelligent multi-agent retrieval.

What it does

• Upload any PDF/document and ask questions about it
• Uses multiple AI agents working in parallel to search and retrieve information
• Rewrites your query for better search results
• Remembers conversation history using summarization
• Self-corrects if retrieved results are irrelevant

Tech Stack

• LangChain + LangGraph — agent orchestration
• Qdrant — vector database for document storage
• HuggingFace Embeddings — semantic search
• Gradio — chat UI
• PyMuPDF4LLM — PDF to markdown conversion

Architecture

User Query → Query Rewriting → Parallel Agent Retrieval → Result Validation → Context Aggregation → Final Answer

How to Run

1. Install dependencies pip install -r requirements.txt

2. Add your documents to the markdown_docs/ folder

3. Run the app cd project python app.py